China has quietly inaugurated the world's first underwater data centre off the coast of Shanghai, a strategic infrastructure project that uses the ocean as a natural heat sink and eliminates the need for freshwater cooling. The facility, located in the Lin-hang special zone, represents a technological leap in addressing the environmental toll of AI-driven computing. By submerging servers in sealed, insulated cooling systems, Beijing is tackling two of the industry's most pressing challenges: water scarcity and energy efficiency. This move comes as Chinese firms like Hikvision and Huawei deploy edge AI in urban surveillance cameras, enabling real-time video analysis without transmitting data to distant cloud centres. The shift reduces bandwidth demands and accelerates response times, marking a departure from traditional face-recognition systems toward text-based video search capabilities.

Yet the global data centre boom is colliding with local politics. In the United States, more than 200 data centres are planned or under construction in competitive House districts ahead of the midterms, fuelling populist anger over rising electricity bills, water use, and farmland loss. Analysts in Washington note that neither party has a clear strategy to manage the fallout, as campaign ads target the tech industry's growing footprint. Across the Atlantic, European policymakers face a similar dilemma: the continent risks falling behind American and Chinese AI giants unless it accelerates data centre investment, but unregulated competition for energy, water, and land is already straining communities. Observers in Brussels argue that Europe could lead by establishing binding environmental standards for AI infrastructure, turning a liability into a competitive advantage.

Meanwhile, the human dimension of AI's rise remains unsettled. Forecasts about job displacement vary wildly, but the core question is not whether jobs will vanish but which ones are vulnerable. Roles involving routine data processing and pattern recognition face the highest risk, while those requiring creativity, complex problem-solving, or physical dexterity appear more resilient. From Tehran to Tokyo, governments are grappling with how to retrain workers and redistribute the gains of automation. The challenge, as experts in London and Berlin emphasise, is not to slow innovation but to ensure that the infrastructure boom—underwater or on land—serves society rather than destabilises it. The next decade will test whether nations can reconcile the voracious resource demands of AI with the political and environmental realities of a warming planet.